Naturwissenschaften DOI 10.1007/s00114-009-0527-8

ORIGINAL PAPER

What’s the buzz? Ultrasonic and sonic warning signals in caterpillars of the great peacock ( pyri)

Veronica L. Bura & Alan J. Fleming & Jayne E. Yack

Received: 24 September 2008 /Revised: 10 February 2009 /Accepted: 2 March 2009 # Springer-Verlag 2009

Abstract Caterpillars have many natural enemies and, Introduction therefore, have evolved a diversity of antipredator strate- gies. Most research focuses on those strategies (crypsis, Acoustic communication in has been exten- countershading, and warning coloration) targeting visually sively researched, with more than 200 published reports guided predators. In contrast, defensive sounds, although on the subject (Minet and Surlykke 2003). The vast documented for more than a century, have been poorly majority of these studies focus on adults ( and studied. We report on a novel form of sound production— butterflies), where hearing and sound production have chirping—in caterpillars of the common European Great evolved multiple times and function in a variety of Peacock moth (). Chirps are broadband, with social and defensive contexts. Comparatively, little is dominant peaks ranging between the sonic (3.7 kHz) and known about the role of acoustic communication in ultrasonic (55.1 kHz) and are generated by a rapid caterpillars, and there is a conspicuous absence of this succession of mandibular “tooth strikes.” Chirp trains are important life stage in reviews on lepidopteran bio- induced by simulated predator attacks and precede or acoustics (e.g., Scoble 1995; Minet and Surlykke 2003). accompany the secretion of a defensive chemical from There is, however, a growing body of literature suggesting integumental bristles, supporting our hypothesis that these that acoustic communication is widespread in larval sounds function in acoustic aposematism. We propose that Lepidoptera. Vibrational communication mediates mutu- these caterpillars generate multimodal warning signals alistic relationships with ants (DeVries 1991;Travassos (visual, chemical, and acoustic) to target the dominant and Pierce 2000), facilitates territorial interactions (Yack sensory modalities of different predators, including birds, et al. 2001; Fletcher et al. 2006; Bowen et al. 2008), and bats, and invertebrates. functions in predator detection (Meyhoffer et al. 1997; Castellanos and Barbosa 2006). Airborne sound produc- Keywords Caterpillar . Acoustic . Aposematism . Chemical tion in caterpillars on the other hand has received less defense . Ultrasound experimental attention. Brown et al. (2007)demonstrated that a common North American silk moth, Antheraea polyphemus produces short mandibular clicks to warn predators of a regurgitant chemical defense. This has been the only experimental study of caterpillar sound production, even though the phenomenon has been Electronic supplementary material The online version of this article reported for more than a century (reviewed in Brown (doi:10.1007/s00114-009-0527-8) contains supplementary material, et al. 2007). It is believed that sound production is not which is available to authorized users. : : only common in the large superfamily Bombycoidea V. L. Bura A. J. Fleming J. E. Yack (*) (silk and hawk moths) but that there exists a wide Department of Biology, Carleton University, diversity of sound production mechanisms and func- 1125 Colonel By Drive, Ottawa, ON, Canada, K1S 5B6 tions. However, these sounds have not been formally e-mail: [email protected] characterized, and the functions remain unexplored. Naturwissenschaften

The focus of this study is on caterpillars of the Great 02614). Larvae were reared on cuttings of poplar (Populus Peacock moth (Saturnia pyri; Fig. 1a): a large, cryptic sp.) housed in an rearing facility at Carleton that ranges throughout southern , Asia University. All experiments were performed on late instars. Minor, and Northern Africa (Rougeot 1971). During a survey of larval defensive behaviors, we noted that S. pyri Sound production mechanism Larval mouthparts were produced long, repeated, high-pitched “chirping” sounds videotaped using a Sony HDR-HC7 HD Handycam that were distinct from the short clicking sounds reported (Tokyo, Japan) equipped with a Sony ECM-MS957 by Brown et al. (2007). Also, unlike A. polyphemus, S. pyri microphone and a macrolens. Videos were analyzed using did not regurgitate. Previous studies on S. pyri show that iMovie 3.0.3 on an Apple computer. To confirm their they use phenolics and related compounds secreted from involvement in sound production, the outer edges of both scoli (bristle bearing outgrowths) in combination with body mandibles were ablated. Larvae were anesthetized using thrashing as their defense strategy (Deml 2001; Deml and carbon dioxide and mandibles ground with a Vogue Dettner 1993, 1995; Fig. 1b). Interestingly, there was no Professional 6700 nail drill (Woodland Hills, CA, USA) reference to sound production in these studies. We equipped with diamond dental burs. In experimental trials, hypothesize that sound production in S. pyri functions to the edges of both mandibles were ground in succession and warn predators of defensive chemical secretions. Our goals the larvae tested for sound production. Controls were were to characterize these novel sounds and their mode of performed using a smooth bit that did not change the production and to determine whether they likely function in mandible structure. Mandibles from known sound pro- acoustic aposematism. ducers were dissected, sputter coated with gold-palladium, and examined using a JEOL JSM-6400 scanning electron microscope (Tokyo, Japan). Methods Sound recording and analysis Recordings were performed Study organisms Saturnia pyri Schiffermüller eggs were in an acoustic chamber (Eckel Industries Ltd., Cambridge, obtained from wild caught females in Kloten and Mohlin, MA, USA) or in an enclosure lined with acoustic foam. Switzerland (Import Permits #P-2007-03105 and #P-2008- Temporal characteristics of chirp trains were measured from

Fig. 1 a A late instar Saturnia pyri larvae. Scale bar, 1 cm. b Higher right mandible with arrows indicating “track marks” from the opposite magnification of a scolus showing the bristle secretions. Scale bar, mandible. e Scanning electron micrograph of the left mandible 1 mm. c Light micrograph of the mouthparts with arrows showing the showing the serrated outer edge. Scale bars, 0.5 mm mandibles. Scale bar, 1 mm. d Scanning electron micrograph of a Naturwissenschaften the first three trains of 14 recordings using Raven is completed, the larva may or may not switch sides Bioacoustics Research Program 1.2 (Cornell Laboratory of between trains, but switching mandibles was never ob- Ornithology, Ithaca, NY, USA). Spectral characteristics served within a particular train. Both mandibles have were examined from five randomly chosen chirps for serrated outer edges (Fig. 1c–e), and we propose that the multicomponent chirps, and all single component chirps edge of one mandible scrapes against the inner surface of (see “Results”) from five . Sounds were recorded the other, where this is evidenced by distinct “track marks” with a Brüel & Kjær (Naerum, Denmark) 1/4 in. micro- (Fig. 1d). These tracks occur on both mandibles, validating phone type 4939 (grid off), placed approximately 4 cm our observations that at least some individuals are “ambi- away from the head capsule, amplified with a Brüel & Kjær dextrous.” The tracks are slightly textured (Fig. 1e), Nexxus conditioning amplifier type 2690, and recorded suggesting that the outer mandible ridges rub against these onto a Fostex FR-2 Field Memory Recorder (Gardena, CA, inner “teeth” to produce individual components or “tooth USA) at a sampling rate of 192 kHz. Spectra were strikes” within a chirp. produced using a 512-point Fast Fourier Transform (Hanning window). Sound pressure levels were deter- Sound characteristics Temporal patterns of sounds were mined by comparing peak-to-peak voltages to the analyzed from the first three trains of multiple attack trials amplitude of a calibrated pure tone. (42 trains from 14 animals), where attacks were directed towards the posterior body region. Chirp trains ranged from Attack trials Attacks were performed by pinching the 0.05 to 4.65 s (mean 1.70±1.12 s) and contained on with blunt forceps, a technique commonly used to average 5.74±3.45 chirps per train (Fig. 2a). A chirp was simulate an attack by a bird or the mandible bite of a defined as a series of tooth strikes separated by a minimum predaceous insect (e.g., Bowers 2003, Grant 2006). Larvae distance of 100 ms. Chirps occurring at the beginning of a were isolated on sprigs of poplar at least 30 min prior to train tended to be longer (mean duration 67.48±23.15 ms, experimentation. Five attacks were delivered to either the n=21) and contained a greater number of tooth strikes head-capsule or the posterior end of the larva, with (mean 5.38±1.53) than those that occurred at the end (mean approximately 5 s between attacks. In an additional trial duration 9.14±10.17 ms, mean number of tooth strikes set, one pinch was delivered to the posterior end of the 1.29±0.46, n=21). caterpillar. Defensive behaviors were videotaped and Spectral analysis was performed on the first three chirp analyzed as described above to determine (1) the mean trains from five individuals where attacks were delivered to number of chirps in 60 s following the first attack, (2) the the posterior region. Sounds were broadband, with most occurrence of sound production and chemical release energy between 3 and 90 kHz (Fig. 2b, c). The energy throughout the attack sequence, and (3) the temporal distribution of multi- and single-component chirps differed relationship between defensive behaviors. (Fig. 2c), with the former having more energy in the ultrasonic range (median dominant frequency 36.00 kHz, mean dominant frequency 34.45±11.93, 88% >20 kHz, n= 25) and the latter having more energy in the sonic range Results (median dominant frequency 14.63 kHz, mean dominant frequency 21.57±15.19, 70% <20 kHz, n=23). When disturbed, S. pyri caterpillars generated distinct and Sound levels were measured from ten chirp trains audible chirp trains. Caterpillars responded acoustically to a obtained from four late instar larvae attacked from the variety of disturbances, including touching the body, or posterior end. Maximum amplitudes, measured from the agitating the plant sprigs, but the most reliable method to loudest tooth strike in a chirp, ranged from 52 to 66 dB SPL induce signaling was to grasp the larva’s posterior end or (median 57 dB SPL), measured at 10 cm from the source. head capsule (Supplementary material, movies 1and 2), both of which induced different responses (see “Attack” Attack experiments A simulated predator attack induced a section below). variety of defensive behaviors, including directed thrashing, head hiding/curling-in, secretion of chemicals from the Sound-producing mechanism Video analysis and ablation scoli accompanied by a foul smelling odor, and sound experiments confirmed that sounds are produced by the production. Attacks to the head region typically resulted in mandibles (Supplementary material, movie 3). Each chirp is the larva curling its head inward while producing sound and produced by either the right or left mandible (Fig. 1c) releasing chemical from the anterior scoli (Supplementary sliding against the inner surface of the opposite mandible, material, movie 1). In contrast, attacks to the posterior and each chirp train results from repeated movements of region caused the caterpillar to thrash towards the site of one particular mandible against the other. Once a chirp train attack while signaling continuously (Supplementary Naturwissenschaften

production was always accompanied by chemical secre- tion (S+C+). In all five-pinch trials resulting in both sound production and chemical secretion, the first chirp preceded or accompanied the first droplet of chemical significantly (P=0.001, χ2=10.9, df=1; Fig. 3d).

Discussion

Caterpillars have many natural predators and consequently have evolved a variety of defenses. The majority of studies on antipredator mechanisms have focused on those target- ing visual predators (e.g., crypsis, countershading, and visual aposematism). However, it is believed that the focus on visual cues is a reflection of our own sensory bias (Lederhouse 1990), and it is reasonable to assume that caterpillars employ a range of signal types (airborne sounds, solid-borne vibrations, visual cues, and volatile chemicals) to communicate with a diversity of predators. Here, we introduce a novel acoustic signal for caterpillars— mandibular “chirping”—and provide experimental evidence for acoustic aposematism. Individual chirps are generated by stridulation, where- by the serrated outer edge of one mandible slides against the textured ridges or “teeth” on the inner surface of the opposite mandible to produce a series of “tooth strikes.” This is a novel form of sound production in caterpillars. Unlike most examples of stridulation in , Fig. 2 Oscillograms of Saturnia pyri sounds recorded from fifth whereby the two body parts are specifically adapted for instar larvae. a Three chirp trains following three consecutive attacks. sound production (e.g., file and scraper) (Ewing 1989), b Time expansion of the bracketed train from part (a), showing typical multicomponent (closed circle) and single component (open circle) the mandibles of S. pyri are not clearly differentiated. The chirps. The arrow indicates the time of attack. The accompanying variability in the inner surface of the mandibles, and the spectrogram shows the frequency distribution of each chirp. c Power amount of pressure against the two parts may account for spectra of multicomponent and single component chirps taken from the diversity in the number of tooth strikes per chirp. three individuals. Time expanded chirps from (b) are inset for each type We hypothesize that stridulation in S. pyri is an aposematic display, warning predators of a chemical material, movie 2) and producing secretions from both the defense. Although the term aposematism is most often anterior and posterior scoli. During a 60-s period, chirp rates associated with visual warning displays, there is a growing following attacks to the posterior region (24.95±19.04, n=20) number of examples of acoustic warning signals, particu- were significantly higher than those following attacks to the larly when the predator–prey interaction takes place under head (12.20±16.79, n=20; P=0.03 two-tailed t test, t38= low light conditions (e.g., moths warning bats: Hristov and 2.25; Fig. 3a, b). Conner 2005, Ratcliffe and Nydam 2008; bees warning Overall, both acoustic signaling and chemical release mice: Kirchner and Röschard 1999) or when sound is part increased with the degree of disturbance (Fig. 3a–c). of a multimodal display to reinforce learning (Rowe and During a 60-s period, larvae attacked five consecutive Guilford 1999). Several predictions satisfy the acoustic times signaled significantly more (24.95±19.04 chirps) aposematism hypothesis. First, sound production is strongly than did those attacked once (3.50±2.78 chirps; P=0.0006 associated with a physical disturbance simulating a predator two-tailed t test, t30=3.85). The relationship between sound attack. Second, the number of acoustic signals generated is production, chemical secretion, and number of attacks is significantly correlated to an increase in the number of illustrated in Fig. 3c. Following a single attack, some attacks. A third prediction is that natural predators of S. pyri caterpillars generated sound without secreting chemical should be capable of hearing these sounds. Saturnia pyri (S+C−), but when attacked multiple times, sound chirps resemble other insect disturbance sounds (Masters Naturwissenschaften

Fig. 3 a Oscillograms showing typical responses to one-, a b five- (head), and five- 44 (posterior) pinch attack trials, with arrows indicating when larvae were attacked. b Mean number of chirps produced over 33 60 s following one-, five- (head), and five- (posterior) pinch attack trials. c Behavioral responses to increasing levels of 22 disturbance (one and five pinch attacks, to the posterior region). S+C+ both sound and chemical 11 production; S−C− neither sound Mean number of chirps in 60 s nor chemical production. d The temporal relationship between sound production and chemical 0 release in five-pinch trials 0 10 20 30 40 50 60 One-pinch Five-pinch Five-pinch Time (s) trials (posterior) trials (head) trials (posterior) c d 100 16

S+C+ S+C- S-C+

75 S-C- 12 Accompanies

50 8 Precedes Percentage of trials

25 Number of five-pinch trials 4

0 0 One-pinch trials Five-pinch trials First chirp precedes First chirp follows or accompanies chemical secretion chemical secretion

1980) in that they are spectrally broadband and temporally generated by stridulating (Masters 1979). A final simple, permitting them to be perceived by a wide range of prediction is that the sounds are associated with an honest predators. But what are the predators of S. pyri? Birds are defense. Our results demonstrate that stridulation most believed to be an important predator (Deml and Dettner often precedes or accompanies scolus secretions, which in 1993), and the sonic components of the chirps overlap with previous studies have been shown to contain compounds the optimal frequency range of most avian predators that function in deterring birds, ants, fungi, and micro- (Schwartzkopff 1955;Dooling1991). The ultrasonic organisms (Deml and Dettner 1993, 1995; Deml 2001). components of S. pyri chirps suggest that they could be Interestingly, sound production is more pronounced during directed towards bats. There is substantive evidence that posterior attacks, corresponding to our observation that caterpillars are common prey of gleaning bats and large more chemical is secreted during these attacks. Overall, our caterpillars can form up to 60% of the diet of some species predictions support the acoustic aposematism hypothesis. (Kalka and Kalko 2006, Wilson and Barclay 2006). Finally, Despite the lack of formal study on the subject of ants are thought to be one of the main predators caterpillar sound production, there is abundant inferential of S. pyri (Deml and Dettner 1995), and although ants do evidence that the phenomenon is widespread in the large not possess ears (Yack 2004), they are sensitive to solid silk and hawk moth (Bombycoidea) caterpillars and borne vibrations (Kirchner 1997). Although we did not test diverse with respect to the mechanisms of sound for vibratory stimuli directly, stridulations may be transmit- production. Earlier reports and preliminary studies have ted through the plant and function as a warning to ants. variously described these sounds as “singing” (Reed Indeed, wolf spiders are deterred by vibrational signals 1868), “the crack of an electric spark” (Bethune 1868), Naturwissenschaften

“crackling–rasping” (Heinrich 1979), squeaking, and click- serotinella (Gracillarioidea Gracillariidae). J Insect Behav 19:1– ing (Yack lab, unpublished). The two formal studies to date 18 Grant JB (2006) Diversification of gut morphology in caterpillars is have revealed two different types of sound production associated with defensive behavior. J Exp Biol 209:3018–3024 (chirping and clicking) that warn of different forms of Heinrich B (1979) Foraging strategies of caterpillars: leaf damage and chemical defenses (toxic spines and regurgitation), repre- possible predator avoidance strategies. Oecologia 42:325–337 senting an interesting example of convergent evolution. Hristov N, Conner WE (2005) Sound strategy: acoustic aposema- tism in the bat-tiger moth arms race. Naturwissenschaften Given the purported ubiquity and diversity of sound 92:164–169 production in this large superfamily (comprising more than Kalka M, Kalko EKV (2006) Gleaning bats as underestimated 3,600 species; Regier 2008), future studies should focus on predators of herbivorous insects: diet of Micronycteris microtis the evolutionary origins, mechanisms, and functions of (Phyllostomidae) in Panama. J Trop Ecol 22:1–10 Kirchner WH (1997) Acoustical communication in social insects. In: different sounds in protecting caterpillars against predators. Lehrer M (ed) Orientation and communication in arthropods. Birkhauser Verlag, Basel Switzerland, pp 273–300 Acknowledgements We thank B. Wenczel and R. Schafroth for Kirchner WH, Röschard J (1999) Hissing in bumblebees: an supplying S. pyri, T. Eberhard for technical assistance, and two interspecific defence signal. Insectes Soc 46:239–243 anonymous reviewers for helpful suggestions to the manuscript. We Lederhouse RC (1990) Avoiding the hunt: primary defenses of gratefully acknowledge research funding by the Natural Sciences and lepidopteran caterpillars. In: Evans DL, Schmidt JO (eds) Insect Engineering Research Council of Canada (JEY), the Canadian defenses. Adaptive mechanisms and strategies of prey and Foundation for Innovation, and Ontario Innovation Trust (JEY). predators. State University of New York Press, Albany, pp 175–189 Masters WM (1979) Insect disturbance stridulation: its defensive role. Behav Ecol Sociobiol 5:187–200 References Masters WM (1980) Insect disturbance stridulation: characterization of airborne and vibrational components of the sound. J Comp Bethune CJS (1868) A musical larva. Can Entomol 1:41 Physiol A 135:259–268 Bowen JL, Mahony SJ, Mason AC, Yack JE (2008) Vibration- Meyhöfer R, Casas J, Dorn S (1997) Vibration-mediated interactions mediated territoriality in the warty birch caterpillar Drepana in a host-parasitoid system. Proc R Soc Lond B 264:261–266 bilineata. Physiol Entomol 33:238–250 Minet J, Surlykke A (2003) Auditory and sound producing organs. In: Bowers MD (2003) Hostplant suitability and defensive chemistry of Kristensen NP (ed) Handbook of Zoology, vol IV. Arthropoda: the Catalpa sphinx, Ceratomia catalpae. J Chem Ecol 29:2359– Insecta. Lepidoptera, Moths and Butterflies, vol 2. WG de 2367 Gruyter, New York, pp 289–323 Brown SG, Boettner GH, Yack JE (2007) Clicking caterpillars: Ratcliffe JM, Nydam ML (2008) Multimodal warning signals for a acoustic aposematism in Antheraea polyphemus and other multiple predator world. Nature 455:96–99 Bombycoidea. J Exp Biol 210:993–1005 Reed EB (1868) A musical larva. Can Entomol 1:40 Castellanos I, Barbosa P (2006) Evaluation of predation risk by a Regier JC, Cook CP, Mitter C, Hussey A (2008) A phylogenetic study caterpillar using substrate-borne vibrations. Anim Behav 72:461– of the 'bombycoid complex' (Lepidoptera) using five protein- 469 coding nuclear genes, with comments on the problem of Deml R (2001) Influence of food and larval age on the defensive macrolepidopteran phylogeny. Syst Entomol 33:175–189 chemistry of Saturnia pyri. Z Naturforsch 56:89–94 Rougeot P-C (1971) Les bombycoïdes (Lepidoptera–Bombycoïdea) Deml R, Dettner K (1993) Biogenic amines and phenolics characterize de l’Europe et du Bassin Méditerranéen. Masson et Compagnie the defensive secretion of saturniid caterpillars (Lepidoptera: Éditeurs, Paris ): a comparative study. J Comp Physiol B 163:123– Rowe C, Guilford T (1999) The evolution of multimodal warning 132 displays. Evol Ecol 13:655–671 Deml R, Dettner K (1995) Effects of emperor moth larval secretions, Schwartzkopff J (1955) On the hearing of birds. Auk 72:340–347 hemolymph, and components on microorganisms and predators. Scoble MJ (1995) The Lepidoptera. Form function and diversity. Entomol Exp Appl 76:287–293 Oxford University Press, New York DeVries PJ (1991) Call production by myrmecophilous riodinid and Travassos MA, Pierce NE (2000) Acoustics, context and function of lycaenid butterfly caterpillars (Lepidoptera): morphological, vibrational signaling in a lycaenid butterfly-ant mutualism. Anim acoustical, functional, and evolutionary patterns. Am Mus Behav 60:13–26 Novitates 3025:1–23 Wilson JM, Barclay RMR (2006) Consumption of caterpillars by bats Dooling RJ (1991) Hearing in birds. In: Webster D, Fay R, Popper A during an outbreak of western spruce budworm. Am Midl Nat (eds) The evolutionary biology of hearing. Springer-Verlag, New 155:244–249 York, pp 545–560 Yack JE (2004) The structure and function of auditory chordotonal Ewing AW (1989) Arthropod bioacoustics; neurobiology and Behav- organs in insects. Microsc Res Techniq 63:315–337 ior. Cornell University Press, New York Yack JE, Smith ML, Weatherhead PJ (2001) Caterpillar talk: Fletcher LE, Yack JE, Fitzgerald TD, Hoy RR (2006) Vibrational acoustically mediated territoriality in larval Lepidoptera. P Natl communication in the cherry leaf roller caterpillar Caloptilia Acad Sci USA 98:11371–11375